The importance of postevent insider trading

MISSING IMAGE

Material Information

Title:
The importance of postevent insider trading evidence from earnings announcements
Physical Description:
vii, 46 leaves : ill. ; 29 cm.
Language:
English
Creator:
Garfinkel, Jon A., 1960-
Publication Date:

Subjects

Subjects / Keywords:
Insider trading in securities   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 44-45).
Statement of Responsibility:
by Jon A. Garfinkel.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002027496
notis - AKL5095
oclc - 33003064
System ID:
AA00003212:00001

Full Text













THE IMPORTANCE OF POSTEVENT INSIDER TRADING:
EVIDENCE FROM EARNINGS ANNOUNCEMENTS












By

JON A. GARFINKEL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA 1\ P\ARTI.A FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


] ()()4

































This dissertation is dedicated to the memory of my maternal grandfather. Wyvnn E.
Walker, whose wisdom and love of finance are the driving force behind these endeavors.








ACKNOWLEDGEMENTS

I would like to thank my committee members Chris James (Chair), David Brown,

M. Nimalendran, Mark Rush and Mike Ryngaert. I would also like to thank Anwer

Ahmed, Matt Billett, Mark Flannery, Joel Houston and Eddie O'Neal for many helpful

and patient discussions, as well as the faculty, graduate students and staff of the

Department of Finance, Insurance, and Real Estate at the University of Florida.

Special thanks go out to my family, without whose encouragement this project

would not have been even imaginable. I would also like to thank my undergraduate

mentors Dana J. Johnson and George Emir Morgan for their continuous support. Finally,

I would like to thank Professor Craig Ruff. who as ; graduate student at Virginia Tech

introduced me to the wonderful world of finance.

I/B/E/S Corp. graciously provided data on earnings ;ind forecasts. This research

was supported in part by a grant from IBM t thrtuuh he University of Florida's Research

Computing Initiative.















TABLE OF CONTENTS




A CKN O W LED G M ENTS................................................... ...................... ...............

A B ST RA C T ................................................................ ............ ..........................

CHAPTERS

1 INTRODUCTION .........................................................................

2 P 'DERS' INCENTIVE TO TRADE AROUND
E,LRNINGS ANNOUNCEMENTS............ ....................

3 DATA AND METHODOLOGY..................................................

3.1 Insider Trading M easures............... .................. ..........
3.2 Abnormal Return Calculation Methodohlov......................

4 RESU LTS...................... ... ........... .... ...........................

4.1 Differential Trading Results ................. ...........
4.2 Differential Profit Results..................... ........ .......
4.3 Results on The Private Information llypotlhesi.............
1.4 Analysis of Types of Intoimaltion twit
Prevent Trading is Based Upon..................... ...........
4.5 Determinants of Benchmark Tradin ..................... .....
-.6 Results on The Fear of Sanctions Hlypotihesis.................
4.7 Firm Policies on Insider Tr.ldin : ind
Fears of Sanctions Results..................... .... ..........
4.8 Robustness of Results \w hen TriadinL DuIrine
the "C rash" of 19 7 is I nored.................. ......... ..........
4.9 Private Information Results and Fiscal Year
End versus Interim Oua;rter Effects.................... ..........

S CONCLUSION.................................


PAGE

iii

vi
















13


13
7
21

24


')
1 1














-12





42
42














REFER EN C ES........................................................................................................... 44

BIOGRAPHICAL SKETCH.................................................................... 46














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE IMPORTANCE OF POSTEVENT INSIDER TRADING:
EVIDENCE FROM EARNINGS ANNOUNCEMENTS

By

Jon A. Garfinkel

August, 194

Chairman: Christopher M. James
Major Department: Finance. Insurance and Real Estate

This dissertation examines the importance of insider trading subsequent to earnings

announcements. Although the extant literature on insider trading has focused on how

insiders behave prior to various corporate announcements. little research has been devoted

to examining the extent to which insiders trade afterr these announcements. This

dissertation fills that gap.

I find evidence consistent with insiders view\in,. po)stevent trading a; s ;in important

option. Specifically, insiders trade more treqtuentlv annd they trade mnore shares following

earnings announcements than at other times. Furthrmiorte. insiders earn significantly

larger profits on postevent trades tha n o othlr trades.

I also find e idence that post e;irnings announcement insider trading is correlated

with several proxies for earnings information. Specific:lly. postevent insider trading is

decreasing in proxies for earnings information that has recently arrived. Furthermore,


I








postevent insider trading is increasing in the difference between one quarter ahead

earnings and current earnings. Finally, insiders tend to buy more (and sell less) following

market underreactions overreactionss) to positive (negative) earnings news.

The observed importance of postevent insider trading can be partially attributed

to insiders being less concerned with SEC sanctions on this trading than on prevent

trading. In particular, higher expected costs of insider trading (associated with a change

in the regulatory climate) led to an increase in postevent insider trading activity relative

to prevent insider activity. Insiders further responded to the change in regulatory climate

by changing the types of information they trade upon prior to earnings announcements.

Specifically, insider buying is more likely to precede negative earnings surprises after the

change in regulations, than before. Finally. insiders earn significantly smaller profits on

sales transactions consummated after the change in regulations.

My results have implications tor the nmeasured incidence ;ind profitability of

insider trading. Specifically, studies that tocus sollyv on insider trading prior to corporate

news events mayv be systematically underestimaruln the incidence ;nd profitability of

insider trading.














CHAPTER 1
INTRODUCTION


A number of studies provide evidence that corporate insiders trade on private

inside information that will soon be revealed to the market.' The common theme

underlying the analysis in these studies is that insiders are trading on an information

asymmetry that exists prior to the corporate news announcement. These studies do not

address the possibility that insiders may use their (assumed) information advantage to

trade profitably after the announcement.

Recent evidence on the market's response to earnings announcements suggests that

insiders may still possess an information advantage over outsiders a after earnings have

been reported. In particular. Bamber [1987] documents increased trading activity in a

firm's stock immediately after a qau;ererlv earnings :nnouncement.- This increased

volume is consistent with a lack of consensusI re :gaUrding, tirm \ value iand thus suIggests an

information asvmmetrv remains after the earnings report. Furthermore, Bernard and

Thomas [1990)] find that the market tails to fully incorpr;ate the implications of current




'For example, Karpoff and Lee [1991] find evidence )t insider selling prior to
seasoned equity offers: Gosnell. Keown and Pinkerton [1)02] document increased insider
selling activity in the five months preceding a;inkruptcy tilings; :ind Lee. Mlikkelson and
Partch [1992] show that insiders increase their buying and reduce their selling prior to
stock repurchases by tender offer.

Beaver [196S] provides similar evidence tor ;nnu;a cIarnines a announcements.








"

earnings for future earnings (i.e. the market underreacts to earnings). Specifically, firms

with the biggest favorable (negative) earnings surprises experience significant positive

(negative) abnormal returns at the subsequent earnings announcement.' If insiders

understand the implications of current earnings for future earnings (or firm value) better

than the market, they can trade on this information for profit after current earnings have

been announced.

This paper examines how and why insiders time their trades around earnings

announcements. Specifically, I focus on the incidence of insider trading after earnings

announcements to highlight the fact that insiders may profit from their information

advantage in ways that were previously ignored by the literature. First, insiders may use

their private information regarding forthcoming earnings to garner a better price on an

expected trade. For example. a planned insider s le will he more profitable following a

positive earnings announcement than preceding one. Second. insiders m:y profit from

private information regarding firm value iater e rnings have been announced. (See

evidence from Bambher [I1 7]. Beaver 11'Q)(S ;and Becrnarii ;jd i Thomias (I )WO].)

I find that insiders are more likely to trIde in the 30 calendar d;ias following an

earnings announcement than at other times. I ;rlso find that insiders trade more (shares)

in the 30 calendar days follow ing an earnings release th;in :it other tines.' Finally.



In other words, if the market full understands currents earnings then they (Bernard
and Thomas) should he unable to predict future earnings responses based on past earnings
figures.

4"Other times" are the 30 calendar days preceding ;In earnings a nnouLncement and the
30 calendar day control (benchmark) period between the pre announcement window and
last quarter's post announcement window.


I










insider trading after earnings announcements is significantly more profitable than insider

trading at other times. This evidence strongly suggests that insiders view postevent

trading as an important option. It also ,ULggcsLt that studies which focus solely on insider

trading prior to corporate news announcements may underestimate hoth the incidence and

profitability of (news related) insider trading.

I find that postannouncement insider trading is related to firm specific earnings

information. Specifically, insiders tend to buy more and sell less in the post earnings

announcement period when one quarter ahead earnings exceed current earnings. They

also tend to buy more when the market underreacts (overreacts) to positive (negative)

earnings news. Finally, insiders use their private information regarding tfrthcoming

earnings news to time their trades and garner better transaction prices. In particular,

insiders increase their selling and decrease their buying after more positive earnings

surprises and after larger stock price run-ups that precede e:lrnin,',s announicements.

Given the observed frequency and intensity of information INsed postevent trading,

I address the determinants of the timing of insider tr:ids. In pr uiu ar. glicn their clear

information advantage prior to earnings annclouncelments. \whv do insiders choose to trade

after them? My evidence suggests that insiders are less concerned \vith sanctions on

postevent trading than on other period trading. Specifically. I examine insider trading

around the passage of the Insider Trading andl Securities Fraud Enforcement Act.

Sevhun [199,21 finds evidence that insiders responidcd t the \cLt hy exccutling fewer

prevent informational trades. I find that insiders execute more postevent trades and

fewer prevent (and benchmark period) trades than \\wuld he expected if the Act had no








4

impact. I also find that insiders earn significantly larger profits on sales executed prior

to this legislation. Finally, 1 find that insider buying in front of negative earnings

surprises is more likely after the passage of this law.

The rest of the paper is organized as follows. Chapter 2 discusses insiders'

incentives to trade around corporate earnings announcements. The primary testable

implications of this research are derived in this section. Chapter 3 discusses the data and

methodology. Chapter 4 presents my results. Chapter 5 concludes.














CHAPTER 2
INSIDERS' INCENTIVES TO TRADE AROUND EARNINGS ANNOUNCEMENTS



For each quarterly earnings announcement (on day 0) I assume there exist three

30 calendar day windows during which insiders can trade: the preannouncement window

[-31,-2], the postannouncement window [1,30] or the benchmark period [-61,-32].' These

three windows are distinguished by the potential information asymmetry that exists

between insiders and outsiders. The insider's choice of which window to trade in

depends on his information advantage as well as the potential costs of trading in a

particular period.

Insiders are assumed to have an information advantage with respect to forthcoming

earnings during the preannouncement window.' Ccteris pa rilus. this information

advantage will encourage insiders to trade prior to earnings announcements. However.

this same information advantage with respect to forthcoming earningss can he used to plan

postannouncement trades. In particular. private knowledge of a forthcoming favorable

earnings announcement will encourage insiders to postpone planned sales of shares until




'I argue that the use of three 30 calendar day windows around an earnings
announcement is appropriate since my proxy for expected earnings (analysts' forecasts)
precede the earnings announcement by approximately 30 calendar days.

'This is consistent with evidence from the extant literature on insider trading (e.g.
Karpoff and Lee [1991] and John and Ling [1991]).








6

after the earnings announcement is made. Thus insiders may benefit from their private

information through either preannouncement, benchmark, or postannouncement trading.3

Insiders may further benefit from private information by trading in the

postannouncement period on the market's underreaction or overreaction to earnings. In

particular, the evidence from Bamber [1987] and Bernard and Thomas [199()] suggests

that an exploitable information asymmetry may exist following the most extreme quarterly

earnings announcements. This represents an additional mechanism by which insiders may

profit from private information in the postannouncement period.

The above suggests that exploitable information asymmetries in existence after

earnings announcements provide insiders with additional incentives to trade postevent as

opposed to prevent. An implication of this is that more trading will be observed in the

postevent period than in the prevent period :mnd benchmark period). This is the central

prediction of The Differential Trading IHypothesis.

The above also suggests that postannouncement insider trading will be correlated

with proxies for recently released earnings info rm;ltion ;ind or future earnings. The

Private Information Htypothesis predicts that ins riders \ ill be more intense sel lers (buyers)

following more favorable (negative) unexpected earnings informal tion releases and after

current earnings that. cx-post. exceed (:ire lower than) uhbsequctnt :earnings. Finally,

insiders are expected to be more intense huyers i sellers) tollo\wing earnings

announcements that the market interpreted too nel; ti\ely (positively).




'Since the benchmark period also precedes the earnings announcement, insiders may
be able to trade profitably on private earnings information during this period.








7

Given the additional incentives insiders have to trade postannouncement (due to

information advantages) the question now becomes, what are the potential costs of trading

postannouncement versus at other times. I suggest that the potential costs of trading

postevent are lower than those associated with trading ;it other times. In particular,

potential SEC sanctions on illegal insider trading represent an important deterrent to

trading on inside information. If such sanctions are less likely on trades consummated

after corporate news announcements than on other period trades. insiders will have an

additional incentive to trade postevent.

The Fear ofSanctions Hypothesis suggests th;t insiders ire indeed less concerned

with sanctions on postevent trading tha oon their period trading. An implication of this

hypothesis is that an increase in penalties on illegal insider trading will imply a smaller

increase in the expected costs of postevent trading than on other period trading and

therefore a shift away from the latter towards tle trmeri. lhe passage of the Insider

Trading and Securities Fraud Enfj)rcementI Act (ITSFEA) increased the penalties on

illegal insider trading ;ind thus provides us \\lth itl iq uIc opportunity o test this

hypothesis. The Fear of Sanctions Hypothesis predicts that insiders will increase their

postevent trading relative to prevent and benchmiark ptricod trading after ITSFEA.

The next section discusses mv data and methodolo,,.














CHAPTER 3
DATA AND METHODOLOGY

My basic sample consists of 25,239 quarterly earnings announcements by 3.816

firms over the period January 1984 through March 1991. This sample met the following

criteria. 1) The quarterly earnings announcement date is available from PC-Compustat

Plus: 2) there exists IBES consensus forecast information for that quarterly earnings

figure: and 3) there were no contemporaneous dividend change announcements associated

with the earnings announcement.

Earnings forecast data and actual earnings per share figures come from the IBES

tapes. The median consensus forecast from the month before the earnings announcement

proxies for earnings expectations. O'Brien 1[ ] 98 finds that this measure dominates time

series based forecasts of earnings. Earnings surprise is defined ;s.

Surprise =( A,-F,)/Price,_,I ( )

where A, is the actual earnings figure for quarter t. :and F: is the median I ES forecast for

quarter t earnings. I scale the forecast error by the price 2 d~ s prior to the

announcement date reported in the Wall Street Journa;l.






The issue of whether to scale 1w the price 30) day;s prior to the earnings
announcement date deserves some discussion. Given that my cross sectional tests control
for the net of market return over the 3( days preceding the earnings report, there should
he little concern over the use of Price


1








9

I also construct a measure of actual earnings growth,

Growth = (A,,+-A,)/Price, 2dy~ (2)

where A,, is the actual earnings figure for quarter t+ and all other variables retain their

prior meanings. All returns and price data come from the CRSP and NASDAQ tapes.

3.1 Insider Tradinc Measures


Insider trading data come from The Ownership Reporting System compiled by the

Securities and Exchange Commission. I use all open market trades by Officers and

Directors of the firm.

Based on the discussion in Chapter 2 I collect insider transactions over three

separate 30 calendar day windows: the prevent period [t-31.t-2]. the postevent period

[t+l.t+30], and the benchmark period [t-(1.t-32]. where t is the iernings announcement

date.- The use of equal sized windows will allow\ me to compare both the %olume and

profitability of insider trading across windows.

I also construct Insider Sale and Purchaise Indexes ilSPIs) fr use in nmy cross-

sectional tests, consistent with previ ous studies of insider tr:adinc. (Sec f or example John

and Ling [191], Damodaran and Liu [1003] ;ndl SVyhun [10 )].) The ISPI is designed




`I do not include trading during the dai; s -1.l] since this is the announcement
window and I don't know whether trades in this ,window occur beIfore or after the
earnings announcement.

'It is important to note that each 30) caiendoar da;v \\indo\w aroundd an earnings
announcement contains 22 (transaction) days on which h a:n insider can consummate a
transaction. Furthermore. insider trading profits can only bh calculated over transaction
days. I therefore calculate my measures of insider proftita ilitv over transaction days.
This calculation is presented in Chapter 3.2.








10

to capture the preponderance of buying or selling behavior in insiders' transactions. The

index is constructed by netting out the number of shares sold by insiders from the number

of shares purchased by insiders in the month of interest, and then dividing by the total

number of shares transacted in by insiders.

ISPI = (Number of shares purchased Number of shares sold)
(Number of shares purchased + Number of shares sold) (3)

The advantage of this measure of insider trading is that it does not suffer from large firm

biases that would skew unsealed measures of net huying/selling behavior.

I construct rmv indices ,of insider trading usine ioth the number of shares traded

and the number of trades executed by insiders.' My index of prevent trading is the ISPI

calculated using trades in the prevent window (henceforth known a;s the earnings month).

Similarly, my index of postevent trading is the ISPI callculated ver, the postevent window

[1,30].

Recent studies of insider trading have noted that some level of trading by insiders

occurs normally ;ind might therefore be considered "benchmark" tradiin. (See for

example Lee. Mikkelson and P;irtch [1W 2]). I contlrl tor the possihility that insiders will

tend to be net buyers or sellers on average, bv including, ;n index of Ienchmairk trading

as a regressor in mv cross sectional tests. My benchrmirk index is the ISPI calculated

over the benchmark period."



'For brevity. I discuss only the shares based measures here. FTrades bh;sed measures
are constructed in a parallel manner.

'To control for possible seaisonalities in insider tr;iding (perh:ips due to tax timing),
the actual benchmark trading index is calculated ;is the average ISPI over the four
benchmark periods preceding the earnings announcement of interest. Specifically, given








11

3.2 Abnormal Return Calculation Methodolohv


I use the basic methodology of Mikkelson and Partch [1986] (with one adaptation)

to assess the significance of abnormal returns earned on insider trades." I assess the

profitability of insider trading over the two month window surrounding the insider's

transaction ([-21,22] in transaction days). If an insider buys shares he profits from a

runup in stock price following the purchase. He also benefits from a decline in the stock

price prior to the purchase. This implies that individual daily abnormal returns should

he calculated as follows:


AR=(R,,,,-Ri,)W for t=[-21.-l] where t)=transalction date

AR=(R. -R,)W tor t=[1.22] where ()=transaction date (4)



where Rj, is the return to stock i on day t. R.,, .-, the contempor aneous market return, and

\ takes on the value of 1 for purchases. -1 for sales.

Tests of statistical significance are h;ised on stiandJarizd tl bniirma. rl-.-Itur (SARs).


SAR,=ARJS\ where (5)




a: December 31. 1988 earnings a;nnounlcementl total Ik-nchn.lark trading, would be
calculated over the months of February. M.\av. A\uLuIt and No\ienmblr rf IO.N8. and then
divided by four to get average monthly benchmark trading.

"A concern with measuring profitability using market model prediction errors is that
estimated alphas may be biased low (high) prior to insider purchases saless. Seyhun
[1986] notes that insiders tend to purchase shares following significant declines in the
firm's stock price relative to the market (implying the negative alpha). I construct my
measure of insider profitability by simply subtracting the market's return from the
observed raw return and cumulating this over the 22 transaction da;vs following the trade.












S(-[1+-L 2 ])1/2 (6)
200 200
S(R,,. -R,)
i=1

V2 is the variance of firm j's abnormal returns over the window [-250.-51]. The

individual daily standardized abnormal returns are assumed to be distributed student's T:

as is the sum of them over 44 days divided by 1/44.

I assess the profitability of prevent and postevent insider trades (separately) by

averaging the 44 dav standardized profitability measure across all N pre (and separately,

post) event transactions, and then multiplying by the square root of N. Since each

standardized profitability measure is distributed normal (0,1), the average will be normal

(0,1,N) and multiplying by IN will result in a standard normal univariate.














CHAPTER 4
RESULTS

4.1 Differential Trading Results


Table 1 contains descriptive statistics on preannouncement. postannouncement and

benchmark insider trading. The percentage of earnings announcements preceded by

insider trading is 20%. while the percentage of earnings announcements followed by

insider trading is 32%. These percentages are significantly different from each other at

the 5% level using a normal approximation to the hinomial test. This result is consistent

with the argument that insiders are more likely to trade after an earnings announcement

than before. I also find that insiders are more likely to trade in the postevent month than

in the benchmark month. Of the 25,239 earnings announcements in my sample. 27.6%

of them exhibited insider trading in the benchmark month. This percentage is

significantly smaller tha:n the percentage t earnings announcements with at least one

episode of postevcnt trading.

I compare the relative intensities as measured bv volumee of shares traded, number

of trades executed. and number of shares per transaction of preannouncement,

postannouncement and benchmark trading. Table I shofxs that mean (median) pre and

postannouncement share volumes are 12.471 (2.400)) and 1).232 (3.70(). respectively.

Chi-square and F-tests reveal that postannouncement insider volume is significantly larger

than preannouncement insider volume at the 1% level. Furthermore, the median number













.. .-* "
- -t ',
[C r'o -; (~~ C) N *

" N !7 -t (N -- r





i Cl
c





if) cc













> I r I" +





r m I
- *



IN
'- NL (-i N -+




*






I r) i



-

L- -1 -







^ ^ ^ri C



-zi








A r A -
1) 1) 0
'r r
-l A f ?r
Hl c'
I____ __

__________________ I-


V






S -
kw






-U 'U
-u






"" -3 = 3





c_ _L 5_
- ..







.- .-- ;_ .
u u .







FE
3^1









,-z
)



























.- .-^ ^
- : "- U .- -
J' .f : yf _
ii- 11
33 3-



















-U <*

ii -U
>- JI










U = = -U -
- ^ ,' < -s -* '
~~ A1
z) _





















z- 'n U- 'J') --
U II *5, -LSU
--U V 3--U
> i 2 U
Ii




-727
) C3 -








15

of shares per transaction is higher for postevent trading than for prevent trading

(X2'=120). Finally, postannouncement trading volume (proxied by the number of trades

and/or the number of shares traded) is significantly larger than the volume of trading in

the benchmark period. The mean (median) number of benchmark trades is 2.33 (1.00)

while the mean (median) volume of shares traded in the benchmark period is 16,451

(3000). Chi-square and F-tests indicate that these value are significantly smaller than the

corresponding postevent values.

To assess whether this difference between postevent ;nd other period trading

volume is due to buying or selling behavior. 1 compare the distributions of pre. post and

benchmark buying volume ;nd the distributions of pre, post and benchmark selling

volume. The results in Table 1 indicate that the difference between postevent and other

period trading volume can be :ittribhuted to ; greater tendency to sell after earnings

announcements than ;t other times. The mean (median) preannouncement,

postannouncement and benchmark period number of shares sold are 9,497 (0()0), 16,868

"2000) and 13,988 (200((). respectively. There is no difference between the mean values

of prevent. postevent and benchmark buying volume.

Further evidence that selling behavior is driving the difference between pre and

postevent insider volume can be found in Figure 1 (see next page). The graph of insider

selling volume closely mirrors the griph of tot;ll insider volume. Also consistent with the

evidence that differences in selling behavior drive the observed differences in insider

volume, I find that the mean and median postevent ISPIs ;ire significantly more negative

than the corresponding prevent measures.













3500


3000 -


2500


S 2000


1500


1000


500



-44 -39 -34 -29 -24 -19 -14 -9 -4 1 6 11 16 21
Transaction Day Relative to Earnings

Market Volume Inside Volume(xl (0)
Insider Buys(xl00) g Insider Sales(xl()O)













Figure 1

Volume of Insider (xl()() and Total Trading
By Transaction Day Relative to Earnings Announcement Date








17

There is no difference between the mean values of postevent and benchmark period ISPIs.

To summarize, postannouncement trading occurs more frequently and is more

intense than preannouncement and benchmark period trading. Although these differences

are driven by insider selling, which is more likely to be liquidity trading, 1 show below

that postevent trading is significantly more profitable than other period trading, which is

consistent with postevent trading being information based.

4.2 Differential Profit Results


I assess the profitability of pre and postevent insider trading using the

methodology outlined in chapter 3.2. Table 2 reports my results. Preevent. postevent and

benchmark insider trading are all significantly profitable. This evidence is consistent with

prior evidence from the insider trading literature.

The results also indicate that postevent trading is significantly more profitable than

other period trading. The mean 44 day net of market returns on pre and postevent trades

are 3.1% and 4.)% respectively. Furthermore, the mean 44 d;iv net of market return on

postevent trades (4.0- .) is significantly larger than the mean 44 day net of market return

on benchmark period trades (3.6%). Postevent median profit measures are significantly

larger than other period median profit measures. These results are consistent with the

notion that postevent insider trading is information based and that it is viewed by insiders

as an important option. Table 3 reports the average profitability of insider trades for each

transaction day around an earnings announcement. The results indicate that each

transaction day shows significant average insider profitability. Figure 2 illustrates how

the profits reported in Table 3 are based on buying versus selling behavior.





















~5 -c
.E


e-
U S






















S ,-
z


















































C- -
u

































-r
~- \- r



=-L_



-T ?

r
1 1
L^ "^ "






t}]

"Z_

-i >






a- a


- ~3
- =









pj -











- p













-i -





-- 5


-I








Table 3
Cumulative 44 (transaction) Day Net of Market Returns on Insider Trades
by Transaction Day Relative to Earnings Announcement Date


BENCHMARK TRADES


Day


Num


Return


PRE EVENT TRADES
Day Num Return


POST EVENT TRADES


Day


Num Return


-43 69 0.05469 -22 542 0.04446 1 943 0.0442 *

-42 391 0.05149 -21 513 0.02755 2 1376 0.0497 *
-41 609 0.03135 -20 556 0.03991 3 1323 0.0455 *
-40 726 0.03563 -19 545 0.04401 4 1145 0.0383 *

-39 702 0.02926 -18 517 0.03455 5 1098 0.0392 *
-38 741 0.04295 -17 524 0.03142 6 1025 0.0317 *

-37 720 0.03014 -16 552 0.03275 7 1034 0.042 *

-36 760 0.03334 -15 485 0.02351 1012 0.0439 *

-35 734 0.04108 -14 498 0.02408 0 1(24 0.0388 *

-34 726 0.0369 -13 464 0.02942 10 995 0.034 *
-33 702 0.04147 -12 517 0.02525 11 965 0.0396 *
-32 659 0.03971 -11 449 0.03234 12 S99 (0.0336 *

-31 644 0.03286 -101 440 0.01867 13 63 0.0360 *
-30 637 0.03505 -) 450 (.02295 14 828 0.0308 *

-29 667 0.03525 -8 40)9 ).01888 15 851 0.0459 *
-- --?----- -- --- I--- -- --- i S--7------7 -
-28 637 0.03246 -7 73 0.0231 i 37 0*

-27 564 0.03418 377 0.02985 17 52 0.0466 *

-26 608 0.03738 -5 353 (.02572 18 762 0.0324

-25 620 0.02924 -4 365 0.05357 19 787 0.0391*
-24 626 0.04118 -3 334 0.03282 i 20 02 0.0416 *

-23 549 0.03513 -2 315 0.04391 l 21 546 0.0359


Day is the number of transaction days prior to (after) earnings announcement that the
insider's transaction occurred on.
Num is the number of insider transactions that occurred on this "Day."
Return is 44 (transaction) day cumulative net of market return averaged across all trades
on this "Day."
"Significant at 5% level.


I

















0.08



0.06



0.04



0.02



0


-0.02


14 19


-43 33 -33 -28 -23 -18 -13 -8 -3 4
Transalction Day Relative to Earnings

SPurchases Sales













Figure 2

Average 44 Day Net of Market Returns
On Insider Purchases and Sales








21

4.3 Results on The Private Information Hypothesis


I examine the relation between postevent insider trading and various earnings

information proxies under The Private Information Hypothesis. Specifically, I relate my

index of postannouncement insider trading (ISPI,,,) to measures of earnings surprise,

earnings growth, the two day abnormal return to the earnings announcement and the

market adjusted return over the 30 days prior to the announcement. My results, contained

in Table 4. control for insiders' "typical" behavior by including the ISPI from the

benchmark pcricid.

Consistent with the notion th;t insiders use their information advantage with

respect to one quarter ahead earnings, I find that insiders increase their buying (selling)

after current earnings announcements that ex post are lower (higher) than subsequent

earnings. The coefficients on the variable comp;aring future earnings to current earnings

in columns two and three (.214,-.18) are signific;int at the 1% level (t=3.33.-2.79).

I also find evidence consistent with insiders trading on the market's under- or

,verreaction to earnings announcements. The coefficient on thie two day a:bnirmal return

is significant in c:ich specification. Controlling tor the effects of earnings surprise.

earnings growth :aHd earnings information that has already arrived, this coefficient

measures the effect mn insider trading of the vati the iri nin he two \ ay alhno~rmal return.

This variation in the market's reaction might he th ught of as a market over or under

reaction to the earnings announcement. The negative coefficient in column one (-1.4.t=-

6.77) indicates that if the two day abnormal return is "too positive" relative to the other

earnings information, insiders increase their selling relative to total trading.








Table 4

Results From Regressing Postevent Insider Trading Index" on Preevent and Benchmark
Indices, Earnings Surprise, Earnings Growth, the Two-Day Abnormal Announcement Return,
the Market Adjusted Return Over the 30 Calendar Days Prior to the Earnings Announcement.


Shares Buying Selling Trades
Variable Indexb Index' Index' Index'

Intercept -.21*** .287*** .550*** -.18***
(-20.2) (51.46) (88.49) (-18.0)
Two Day -1.4*** -.75** .684*** -1.4***
Abnormal Return' (-6.77) (-7.08) (6.537) (-7.10)

Earnings Surprise" -.41"* -.24*** .216"* -.384
(-2.68) (-2.95) 2.742) (-2.54)
Earnings Growth" .354** .214* -.18" .344*"
(2.867) (3.327) (-2.79) (2.832)
Runup' -1.5" -.82*" .796*, -1.5"
(-16.6) (-17.6) (17.22) (-17.0)
Preevent Insider Trading .340*** .249"* .171"* .352"*
Index (17 99) (14.82) (13.02) (18.55)

Benchmark Insider Trading .39" .2 7""* .238," .416""-
Indexk (23.08) (15.24) 20.44) f (24.18)

N 7474 7474 7474 7474

F-Statistic 288** 174'*" 212" 310)"*



IS~
SAdj-R2 .172 .1222 1447 .19 88

"ISPIr,,,,
'Shares based measure of Postevent Insider Triding Index (I SPI ,,i.
Number of shares bought rel;tive to total shares transacted in (Iduring post event period).
dNumber of shares sold relative to t, tal shares transacted in duringc post event period).
"Trades based measure of Postevent Insider Trading Index (ISPI,,).
Two-day abnormal return to the earnings announcement.
(Actual Earnings Median Analyst's Forecast)/Price, .,.,,,
'(Actual Earnings,, Actual Earnings,)/Price, .,
'Stock's Cumulative Net of Market Return over the calendar window [t-31.t-2]. where t is the
earnings announcement date.
'Shares Index,,, Buying Index,., Selling Index ,. Trades Index.
kShares Index,,,,, Buying lndexb,,,h, Selling Index h,,.,, Tr;des n..l\,. ,
T-statistics in parentheses: Significance levels: "**-o1%: -5%; *-1)%








23

I find evidence that insiders use their private information regarding forthcoming

earnings to time their trades and achieve better transaction prices. Specifically, I find that

positive news precedes insider sales and negative news precedes insider purchases.

Postevent insider buying is decreasing in my measure of earnings surprise. The

coefficient of -.24 (column two) is significant at the 1% level (t=-2.95). Similarly,

postevent insider selling is increasing in earnings surprise (coefficient=.216,t=2.74),

indicating that insiders sell more after more positive earnings shocks.

Additionally, postevent trading is decreasing in the () calendar day

preannouncement market adjusted return (runup), which proxies tor information regarding

the forthcoming news that leaked early.' The interpretation of the coefficients on the

runup variable mirrors the interpretation of the coefficients on the surprise variable. More

positive earnings information leakage elicits more postevent selling.

The evidence that insiders wait and trade against the price reaction to earnings

information leakage (the significant coefficients on the runup variables) is also consistent

with evidence from the extant literature in inider. trading. Specificallv, Se\ hun (199())

provides evidence consistent \with insiders correctly identifying undervaluations

(overvaluations) in their firm's stock, and purchasing (se ling) shares at those times. My

evidence is also consistent with this pattern.





Easton & Zmijewski [1989] interpret the 30-day runup variable in their study in
much the same manner. Specifically, they argue that the negative correlation between
their two-day abnormal earnings announcement return and their runup measure is
consistent with a positive association between the stock return prior to the two-day
window and the measurement error in unexpected earnings.








24

Finally, my evidence indicates that net insider buying (selling) tendencies in the

benchmark and prevent periods are associated with net buying (selling) tendencies in the

postevent period. The coefficient on the benchmark ISPI (.396 in column 1) is

significantly positive, as is the coefficient on the prevent ISPI (.34 in column 1).

4.4 Analysis of the Types of Information that
Preannouncement Tradine is Based Upon


I examine the determinants of preannouncement insider trading by regressing my

insider trading measures (for preannouncement trades) on measures of earnings surprise,

earnings growth and the market adjusted return from the trade date through the earnings

announcement (MAR). The results are contained in Table 5.

Controlling for earnings surprise and earnings growth, insider trading is increasing

in the average market adjusted return from the trade date through the earnings

announcement. The coefficient on MAR (.617 in the first specification) is significant at

the 1b% level (t=3.NS). I also find that, controlling for earnings growth and MAR, insider

trading is decreasing in earnings surprise. TFken together. this evidence suggests s that

insiders are trading profitably on earnings information not picked up by proxies for

earnings torecast errors. For example, earnings surprise 1may he negative due to a one

time charge but sales mayv he higher than expected, Clicitin a; positive market reaction.

The evidence suggests that insiders are net buyers prior to this type of event. Finally, the

negative coefficient on earningss surprise ma;iv e di_ e ini p;irt to tfears of ,;nctions on

prevent trading. If insiders can point to the negative relation between their trading and

the earnings surprise they maiv feel safer \vith respect to SEC sanctions.































33(5
(4.00)


Table 5

Results From Regressing Preevent Insider Trading Index" on Benchmark
Index, Earnings Surprise, Earnings Growth, and the Market Adjusted Return from the
Trade Date through the Earnings Announcement Date


Shares Buying Selling Trades
Variable Index index Inndexd Index'

Intercept -.09*** .32** .479*** .07***
(-6.59) (45.50) (58.89) (-5.70)
Earnings Surprise' -.92*** -.47"* .464*'** -.863***
(-3.31) (-3.23) (3.241) (-3.15)
Earnings Growth' -.014 .0198 -.102 .0033
(-.(9) (.25) (-.03) .0)22)


Market Adjusted Return from
trade date through announcement"


Benchmark Insider Trading .026- .437-"" .374*"
Index' (32.41) (22.32) (27.20)

N 4697 4697 4697

F-Statistic 272*** 133"'" 194.".

Adj-R2 .1875 .101 1 .1411


"ISPIp,
'Shares based measure of Preevent Insider Trading Index (ISPI.,,).
'Number of shares bought relative to total shares transacted in (during pre event period).
'Number of shares sold relative to total shares transacted in (during pre event period).
Trades based measure of Preevent Insider Trading Index (ISPI,,).
'(Actual Earnings Median Analyst's Forecast)/Price,,,,.
'(Actual Earnings,, Actual Earnings,)/Price .
hStock's Cumulative Net of Market Return from the transaction dite through the earnings
announcement date.
'Shares IndexL,,h,, Buying Index.,,,, Selling Index,,,.,nh, Trades Index,,.,,


T-statistics in parentheses: Significance levels: ""-1%;: '"-5': *- 1W)


.617***
(3.88)


-.353""
(-4.316)


.646*"*
(4.13)








26

Prevent insider trading is not significantly related to earnings growth as measured

by the difference between one quarter ahead earnings and current earnings. This is not

terribly surprising since insiders can wait to trade on one quarter ahead earnings until just

after current earnings are released, and they will likely he less concerned with SEC

sanctions. I formally test this notion that insiders may be more concerned with SEC

sanctions on prevent trading than on postevent trading in chapter 4.6.

4.5 Determinants of Benchmark Trading


The fact that I use benchmark trading as a control measure does not preclude the

possibility that such trading is correlated with earnings information. Table 6 contains

results from regressing ISPI.,,lnh on measures of earnings surprise, earnings growth and the

market adjusted return from the trade date through the earnings announcement (MAR).

The results in Table ( ;ire very similar to those described in Table 5. Controlling

for earnings surprise and earnings growth, benchmark trading is increasing in MAR

(t=3.64 in column I). Furthermore, benchmark trading is decreasing in earnings surprise

(t=-2.31 in column 1) after controlling for earnings growth and NMAR. These results

suggesttt that insiders trade profitably on earnings information not picked up by forecast

error proxies. :is early :n s tw\\ months prior to the earnings announcement. Finally,

benchmark trading does not appear to be correlated with nm proxy tor earnings growth.








27

Table 6

Results From Regressing Benchmark Period Insider Trading Index" on
Earnings Surprise, Earnings Growth, and the Market Adjusted Return from the
Trade Date through the Earnings Announcement Date


-.1014
(-1.4)


(1.4)


.1014
(1.4)


Market Adjusted Return from .3X82*** .19l**" -.19*" .386js**
trade date through announcement" (3.644) (3.644) (-3.64) (3.710)

N 6469 6469 6469 6469

F-Statistic I6.23*** 6.23"* .23""" 5.70**"

Adj-R' .0024 0024 0024 .0022

ISPI1venh
hShares based measure of Benchmark Insider Trading Index (ISPI.,,.).
'Number of shares bought relative to total shares transacted in (during pre event period).
'Number of shares sold relative to total shares transacted in during pre event period).
'Trades based measure of Benchmark Insider Trading Index (ISPI,,,).
(Actual Earnings Median Analyst's Forecast),Price, .,
(Actual Earnings,, Actual Earnings,),Price, ,,.
'Stock's Cumulative Net of Market Return from the transaction date through the earnings
announcement date.


T-statistics in parentheses: Significance levels: *- I';: -5_ : -1:_(


1.3)








28

4.6 Results on the Fear of Sanctions Hypothesis


An implication of greater fears of sanctions on prevent trading (than on postevent

trading) is that an insider's choice to trade before or after an earnings announcement will

depend on the current regulatory regime (whether ITSFEA has been passed). Given that

ITSFEA increased the penalties on illegal insider trading (see chapter 2), postevent trading

should be higher and prevent trading should be lower (in the post Act environment), than

would be expected in a world where fears of sanctions are equal on both types of trading.

The intuition is as follows.

If insiders anticipate the likelihood and cost of sanctions to be the same whether

trading occurs prior to or after earnings announcements, then the expected sanctions on

either type of trading are equal and remain that wav even after the Act is passed. The

alternate case. in which sanctions are more likely on prevent trading, suggests that the

expected costs of this type of trading are higher after ITSFEA than the expected costs of

postevent trading. This should mitigate insiders' desires to trade prior to earnings

announcements (relative to their desire to trade after earnings announcements).

I test this implication hv relating insiders' trade timing n to an indicator variable for

the current regulatory regime (whether or not ITSFEA had been passed). Specifically,

conditional on insider trading being observed either before or matter ;in earnings

announcement, the announcement is classified as a pre trading announcement if there was

at least one episode of preannouncement trading, else the announcement is classified as

a post trading only announcement. This classification is related to the current regulatory

regime using a contingency table.








2()

The results are contained in Table 7. I find evidence consistent with insiders

fearing sanctions on prevent trading more than they do on postevent trading. In

particular, there are more episodes of postevent trading after ITSFEA than would be

expected if insiders fear sanctions equally on pre and postevent trading. Under the null

hypothesis of no difference in fears of sanctions there are 1912 expected cases of

postevent trading after ITSFEA: 2092 cases are observed. Furthermore, there are 1557

observed cases of prevent trading after ITSFEA: 1737 cases were expected under the

null. The Chi-square statistic on the test of independence between trade timing choice

and the current regulators regime is 12.2,. which is significant at the 1% level. The null

hypothesis of independence is rejected.

Results not shown also indicate that the Act affected the willingness of insiders

to trade during the benchmark period versus the postevent period. Using the methodology

described above, I find that after ITSFEA. insiders increased their tendency to trade

postevent (as opposed to during the benchmark period) more than we would expect if the

Act hld no effect. The measured Chi-squ;are statistic tr 3.S7 is signitic;ant it the 5%

level.











Table 7

Contingency Table Relating the Insider's Trade Timing
(During the Pre or Postevent Period)"
to the Current Regulatory Regime
(Before or After the ITSFEA)h
(Insider Trading and Securities Fraud Enforcement Act)

Test of the Fear of Sanctions Hypothesis

There Exists Post Trading There Exists Pre
and No Pre Trading Trading Totals

Earnings 3437 -- Actual Cases 3465 -- Actual Cases
Announcement
Before ITSFEA (3617) Expected Cases' (3285) Expected Cases' 6902

Earnings 2092 -- Actual Cases 1557 -- Actual Cases
Announcement
After ITSFEA (1912) Expected Cases' (1737) Expected Cases' 3649

Totals 5529 5022 10,551


/2(1) = 12.28***

Test of independence between choice of when to trade and current regulatory regime is rejected
at better than the 1 o level.

'All numbers are "conditiionn onm tr:iding" bv :it least one insider. durin- the aIppropriate period.

I'TSFEA was passed in November 4f 1988.

'Expected cases are based on the assumption of independence between the insider's choice of
when to trade and the current regulatory reeime.

Significance Levels: "'"-1'%








31

I examine whether the results in Table 7 are driven by an increase in buying or

selling behavior in the postevent period relative to the prevent period. Specifically, I

calculate the difference between the number of shares bought (sold) in the post earnings

announcement window and the number of shares bought (sold) in the pre earnings

announcement window. This number is scaled by the total number of shares bought

(sold) in both windows. The resulting Post Minus Pre Buying (Selling) Index is then

related to an indicator variable for the current regulatory regime (before or after ITSFEA),

and a control variable for the total volume of trading in the post earnings announcement

window relative to the total volume of trading in the pre earnings announcement window.

The results are contained in Table S. 1 find evidence consistent with an increase

in selling in the post earnings announcement window relative to the pre earnings

announcement window, following the passage of ITSFEA. In the selling regressions (with

and without the relative volume control), the coefficients on the indicator variable for the

current regulatory regime, .052 ind .(44 respectively, are positive and significant (t-

statistics = 2.32 and I.S ). In other words, inside rs increase their selling behavior in the

postevent period relative to the prevent period, following the pa:ss55ag; o ITSFEA. These

results do not hold for the Post Minus Pre Buving Index. The coefficients on the

regulatory regime indicator variable (.(021 .11 7 are positive but statistically insignificant.

t=.766 and .627). Insiders did not change their postevent buying relative to prevent

buying, following the passage of ITSFEA.








32

Table 8

Results From Regressing "Post Minus Pre" Buying and Selling Indices on
an Indicator Variable for the Current Regulatory Regime,
and the Relative Volume of Total (Insider and Outsider) Trading in the
Post versus Pre Periods.
Tests of the Fear of Sanctions Hypothesis


Post Minus Pre = i of shares huc.ht in post period
Buying Index (# of shares Iought in post period


'Post Minus Pre
Selling Index

"Relative Volume =


- if ithaires Ihouht in pre period)
+ 4 of shares bought in pre period)


(# of shares sold in post period -i of shares sold in pre period)
(4 of shares sold in post period + = of shares sold in pre period)


Total Volume of Shares Tra:ded in Post Period
Total Volum e of Shares Tr:ded in Pre Period


(outsiders -
insiders)


T-statistics in parentheses
Significance levels: -1' : --5(f%: (0%








33

Given the evidence that insiders reacted to ITSFEA by reducing their prevent

trading activity in favor of trading in the postevent period, the question of whether they

also changed the type of information they trade upon arises. For example, the negative

correlation between prevent trading and earnings surprise may he driven by trading that

occurred after the Act. I examine whether the types of information that insiders trade on

prior to earnings announcements changed after the Act in Table 9.'

The first column of Table 9 indicates that insider buying was more likely to

precede negative earnings surprises after the Act. The coefficient on the interactive

surprise variable (which takes on the value of surprise post Act. () otherwise) is -1.23 with

an associated t-statistic of -4.44. Furthermore, the coctficient on surprise (not interacted)

is -.0001 (t=-.51 ). Taken together, this evidence is consistent with insiders changing their

prevent buying behavior after the Act in such ;i \v;I thait is likely to limit SEC sanctions.

In particular, insiders who executed prevent buys tifoowing the Act could simply point

to the observed average negative earnings surprise. should their trading be questioned.

At the same time, this chaiiin e in prevent ib'vinii behavior did not affect the

profitability of these trades. Column 3 to Tible indicates that insider buying is

increasing in runup both before and after the Act. The coefficient on runup is .403

(t=4.16), while the coefficient on runup interacted with POSTITSFEA (=1 after the Act,

0) otherwise) is insignificant. Column 5 indicates th:it the pjroitable information which



Results not shown indicate that there was no change in the types of information that
insiders trade on in the post event period, following the Act.

:For ease of presentation I do not include mv measure of earnings growth in the
analysis. Mv results do not change upon inclusion of earnings growth as a regressor.


I













*= ,-' -0' 2 2 ^ K P
*
00&1 00N N 00 *C 3r- *
O X C rl" C', r,4c x C7 r
5'L b~fn O;
00 rz r tNN Z -






m y
- a
en


>




L1 E 3 i2^"~ ^ l P
v 00 ,- *n






*^ < rS 1~ Cl *^ 'C *. il ~. '* I
LLc ^ 1^ Z\C -; '* 00 0 : 'C 0 *
00'C ~~~~'r r4 rn f't~~j 0 I N C








Cl _s : =
f nt










I *

3 Cl S~- -. *
- r"l '1 ? ;C C1 <. ~ < *
a-r









r_ -I C Cl | ^ T


| B














-L4
I" I` _












It 1 : c'- 1 ;: =1
r-r j rI *
lir 'r ,














'N/ -c r. -t -



-- Cl i K .1 I 1
rr-





















Iy F < = <* ='^ -
": ^ *- -- U ,


I) -: L U <, ;
~d c" I I ~? r

SIr Ir I


K" C~C r
Lr 7 'V
17": __ j


r-
-3 E E







>> = 4









-L. -
SL* .




*"













- >_







S r -
= t'-





















" '- i
L- 5-








35

insiders bought on was uncorrelated with the measured earnings surprise. Controlling for

earnings surprise, prevent buying is increasing in runup. This is consistent with insiders

buying before earnings announcements on positive information that is not captured by

coarse measures like earnings forecast errors.

Table 10 further examines the profitability of trades before and after ITSFEA.

The results indicate that insider profits on sales transactions declined after the Act,

consistent with the notion that insiders became more concerned with SEC sanctions after

ITSFEA. Interestingly, the profitability of insider purchases increased after the passage

of ITSFEA. This appears to contradict the notion that insiders became more concerned

with sanctions ti'llowing ITSFEA. However, insiders may simply believe that sanctions

are less likely to he imposed on their purchases. Future research can address this issue.


1


























.2 ** *^
C C
L <- <^ ^ n^
Cl)L Cl -I .f i
< Q S i ,t [




I i n
L~ .E ~ ~I :
*/!

>___ __ ___












r ._.j
Ut






>, if --
*. Ciii ", r


'\ -' "

I *_'( i
i |

|---r I-,-----






I 5 Ti ri ri /,j m cr^
I u xt ri rl Ir- -


's /i i









~i | ^ I j !



j! | ~ = C '
jl -_ ., I ? I









* I
I ~ ~ ~ ~ ~ Z ~ C r C l. nj

b i.)i Cl)


rrl
t
r^-








'L







Z














/:-
"1










s







r.














r.




J


-)











aJ -















- -


-1 -




1*








37

4.7 Firm Policies on Insider Trading and
Fears of Sanctions Results


It is possible that the firms in my sample instituted substantially restrictive policies

on insider trading following the passage of ITSFEA. If this is the case. my conclusion

that increased federal government scrutiny and penalties directly affected insiders'

behavior may be premature. In particular, the observed switch in insider trading behavior

from more prevent trading to more postevent trading may have been caused by more

restrictive firm policies that arose in response to the Act.

I examine the possibility that changes in firm policies led to the observed change

in insider behavior by analyzing firm specific policies on insider trading both before and

after ITSFEA. For a sample of 25 firms' (16 with at least one episode of prevent

insider trading, nine (9) with no episodes of prevent trading) I collected information on

their policies regarding prevent trading and when these policies were put into place.

Each firm has a policy that at least discourages prevent trading and eight (8) firms

strictly forbid it.

However, of these 25 firms, onl three (3) (t them put these policies into place

after ITSFEA (12%). If we assume that myv sample if respondents is representative, most

firms (88%) made no change in their policies regarding insider trading following the Act.

This suggests that the Act, and not firm policy changes in response to the Act. caused the

observed chance in insider behavior.


50 firms were contacted. Only 25 responded despite follow-up calls on my part.








38

4.8 Robustness of Results when Trading Durini
The "Crash" of 1987 is Ignored


It is possible that trading immediately following the "Crash" of October 1987 was

more information based than other insider trading. Specifically, Seyhun [1990] provides

evidence consistent with insiders being able to identify under and over valuations in their

firm's stock and trade profitably on this information. I test The Private Information

Hypothesis for the subsample of insider trades that did not occur in the month of October

1987 in order to assess the robustness of my Private Information results.

Table 11 presents the results from testing The Private Information IHypothesis

using trades that did not occur during October 1987. The results indicate that trades

during the "Crash" month are not driving my results. Insider trading is still increasing

in one quarter ahead earnings. The coefficient in specification one (.31) is significant at

the 5% level (t=2.4S). Also, postevent trading continues to be negatively related to

earnings information that has recently arrived. The coefficients on the 30-day

preannouncement market adjusted return ;ire :ill i,-nitic;int. Finally. postevent insider

trading is decreasing in the proxy for ma rket c\, rre:;iction (underreaction) to negative

(positive) earnings news (t=-7.74).








Table 11

Results From Regressing Postevent Insider Trading Index" on Preevent and Benchmark
Indices, Earnings Surprise, Earnings Growth, the Two-Day Abnormal Announcement Return,
the Market Adjusted Return Over the 30 Calendar Days Prior to the Earnings Announcement.
Without October 1987 Trades


Shares Buying Selling Trades
Variable Indexb Index' Index' Index'

Intercept -.238*** .270** .569*** -.212***
(-23.02) (48.17) (90.23) (-20.89)
Two Day -1.7*** -.93**" .S6*** -1.7*""
Abnormal Return' (-7.74) (-8.04) (7.51) (-7.94)

Earnings Surprises -.46*** -.26***" .24*** -.43**
-2. S7) (-3. 1 ) (2.97) (-2.74)
Earnings Growth" .309" .19"** -.151" .311"
(2.48) 2.94) (-2. ) (2.53)
Runup' -1.5" -.l" .786" -1.5"*
(-16.5) [ (-17.0) (16.73) (-16.5)
Preevent Insider Trading .3160 .236" .s158 .,328"*
Index' I 1 6.49) ( 13.62) 1 1.96) i 17.04)

Benchmark Insider Trading .403' .272* .237- .424"
Index' 23.35) (16.03) 20.19) 124.47)

N "142 7142 7142 7142

F-Statistic 77' 17,"' -99- ', ))"

Adj-R2 .1 3 .1202 .1441 .2003
'IS P
bShares based measure of Postevent Insider Trading Index (ISPI ,,,,).
'Number of shares bought relative to total shares transacted in (during post event period).
'Number of shares sold relative to total shares transacted in (during post event period).
Trades based measure of Postevent Insider Trading Index ( ISPI ,,).
'Two-day abnormal return to the earnings announcement.
(Actual Earnings Median Analyst's Forecast)/Price, _.,,.
h(Actual Earnings,, Actual Earnings,)/Price,.,,m.
'Stock's Cumulative Net of Market Return over the calendar window [t-t-31t-2] where t is the
earnings announcement date.
'Shares Index,,, Buying Index,,, Selling Index,,., Trades Index,,.
kShares Indexhn Buying lndex,,,,, Selling Index 1.,,, Trades Index,,,,,
T-statistics in parentheses; Significance levels: **-1%; "-5%; "-1 10








40

4.9 Private Information Results and Fiscal Year End
versus Interim Quarter Effects


It is possible that insiders have a differential information advantage around interim

quarter earnings announcements than around fiscal year end reports. Since interim quarter

results are generally unaudited, insiders may he more sure of their information advantage

around these than around audited earnings announcements. On the other hand, firms

generally publish fiscal year end results with a longer lag following the end of the

measurement period than interim quarter results. This could provide insiders with an

opportunity to improve their information advantage over outsiders for these

announcements. The question of whether insiders possess a greater information advantage

and trade on it for interim quarter or fiscal year end earnings is an empirical one.

Table 12 presents results on The Private Information Iypothesis controlling for

fiscal year end versus interim quarter effects. The results indicate that there is some

difference in the correlation between postevent trading and information proxies for fiscal

'car end versus interi quarter r ;announcements. Specifically, the correlation between

postevent insider trading and both earnings growth ;ind runup is stronger for fiscal year-

end earnings announcements than for interim quarter announcements. The T-statistics of

the coefficients on the interactive growth and runup variables (which take on the value

of zero for interim quarter announcements while e retaining their observed va;lute for fiscal

year-end announcements) are both significant. This result is consistent with insiders

possessing a greater information advantage with respect to fiscal \ear end earnings

announcements that they actively trade upon.


















ja




'-C

a.- -
1. 1

U,

H 2 E
_r 3
S u c











UC
U C







-5 >-







S> ": 5





-^ < ^>)
3cE






P~ U
u, -,







-:)* -:'-J) -



'-C,, *r: -
^? ^-2 .

















-1) I- e
^*s = =










U ?1 : 3





-~--~

















'PP
1
^) X 5'*





c L- _



uu~
a;


>..
*
*



r
cn







*
C-



cI,
*
vl







r,,







* <


.4--

'-
t,,
+-
>-

























-i"





LLp
h




f~


4'-



r ,









-t- -

>r













-1
r.-
t












t,,


- I I
71,



.I
t- I


'L





'U 'U














CHAPTER 5
CONCLUSION

The literature on insider trading contains numerous examples of insiders trading

prior to the release of private information. However, few studies have examined the

importance of postevent insider trading. I address this gap using a large sample of

earnings announcements. I find evidence that postevent insider trading is much more

likely than prevent and/or benchmark period insider trading. Postevent insider trading

is also more voluminous than prevent and benchmark period insider trading. Finally,

postevent insider trading is significantly more profitable than both prevent and

benchmark period insider trading.

Postannouncement insider trading appears to be based on insiders' private

information regarding earnings growth. Specifically, larger (smaller) one quarter ahead

earnings elicit more buying (selling) in the postannouncement period. Insiders also appear

to wait until earnings information is revealed to the market l;nd then trade against the

market's reaction. In particular, insiders sell more and buy less following positive runups

in the stock price (prior to earnings announcements) and following positive earnings

surprises. Finally, insider trading appears to be correlated with market overreactions

(underreactions) to negative (positive) earnings news.

I find evidence that insiders fear sanctions less on postevent trading than on other

period trading by examining the behavior of insiders around the passage of ITSFEA (The








43

Insider Trading and Securities Fraud Enforcement Act). Specifically, after ITSFEA

insiders increase their postevent trading and decrease their prevent and benchmark period

trading relative to what we would expect if they are equally concerned with sanctions on

all three types of trading. Insiders further responded to the Act by altering their prevent

buying behavior. Specifically, insiders were more likely to buy before negative earnings

surprises following the passage of ITSFEA than before. Finally, insiders earn

significantly smaller profits on their sales consummated after ITSFEA.








REFERENCES


Bamber, Linda, 1987, Unexpected Earnings, Firm Size, and Trading Volume Around
Quarterly Earnings Announcements, The Accounting Review 62, 510-532.

Beaver, W., 1968, The Information content of Annual Earnings Announcements, Journal
of Accounting Research, Supplement 1968. 67-92.

Bernard, Victor and Jacob Thomas, 1990, Evidence that Stock Prices do not Fully Reflect
the Implications of Current Earnings for Future Earnings, Journal of Accounting and
Economics, 13, 305-340.

Brous, Peter Alan, 1992, Common Stock Offerings and Earnings Expectations: A Test of
the Release of Unfavorable Information, Journal of Finance. 47, 1517-1536.

Collins, D. and S.P. Kothari, 1989, n AnAalysis of Intertemporal and Cross-Sectional
Determinants of Earnings Response Coefficients. Journal of Accounting and Economics,
11, 143-181.

Damodaran, A. and C.H. Liu. 1993, Insider Trading as a Signal of Private Information,
Review of Financial Studies. 0. 79-12(0.

Easton, P. and M. Zmijewski. 1989, Cross-Sectional Variation in the Stock Market
Response to Accounting Earnings Announcements. Journal of Accountinn and Economics,
11, 117-141.

Gosnell, T., A. Keown and J. Pinkerton. 1992, Bankruptcy and Insider Trading:
Differences Between Exchange Listed and OTC Firms. Journal of Finance. 47, 349-362.

Hirschey, M. and J. Zaimna. 1,89. Insider Trading.. Ownership Structure :ind the Market
Assessment of Corporate Scll-Offs, Journal of Finance. 44. 71-'().

John, K. and L. Ling, 1991, Strategic Insider Trading Around Dividend Announcements:
Theory and Evidence, Journal of Finance. 46. 1 361-139().

Karpoff, J. and D. Lee. 1991. Insider Trading Before New Issue Announcements,
Financial Management, 2(, 18-26.

Lee. D., W. Mikkelson and M. Partch, 1992. Mianlagers' Trading Around Stock
Repurchases. Journal of Finance. 47. 1947-1962.

Mikkelson, W. and M. Partch. 1986. Valuation Effects of Security Offerings and the
Issuance Process, Journal of Financial Economics, 15, 31-60.










O'Brien, Patricia, 1988, Analysts' Forecasts ;is Earnings Expectations. Journal of
Accounting and Economics, 10, 53-83.

Penman, S., 1985, A Comparison of the Information Content of Insider Trading and
Management Earnings Forecasts, Journal of Financial and Quantitative Analysis, 20, 1-17.

Seyhun, H. Nejat, 1986, Insiders' Profits, Costs of Trading, and Market Efficiency,
Journal of Financial Economics. 16, 189-212.

Seyhun, H. Nejat, 1988, The Information Content of Aggregate Insider Trading, Journal
of Business. 61, 1-24.

Seyhun, H. Nejat, 1990, Overreaction or Fundamentals: Some Lessons from Insiders'
Response to the Market Crash of 1987. Journal of Finance, 45. 1363-1388.

Sevhun. H. Nejtt. 1992. The Effectiveness of Insider Tiading S;nctions. Journal of Law
and Economics. 35. 140)-12.














BIOGRAPHICAL SKETCH

Jon Garfinkel was born on May 5, 1966, to Patricia and Stephen Garfinkel in

Washington, D.C. He attended the Virginia Polytechnic Institute and State University

(Virginia Tech) from September 1984 through May 1988. achieving a Bachelor of Arts

degree in economics. Following one year of work as an auditor for the United States

General Accounting Office, he enrolled in the Ph.D. finance program at the University

of Florida (anticipated I radutation on August 0. 1994).














I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosoply.


Christophe James, Chairman
Sun Bank Professor of Finance,
Insurance, and Real Estate

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
;s a dissertation for the degree of Doctor of Philosophy.


David T. Brown
Associate Professor of Finance.
Insurance, and Real Estate

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Michael D. Ryngaert
Associate Professor of Finance.
Insurance. and Real Estate

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Mahendrarajah Nimalendran
Assistant Professor of Finance,
Insurance, and Real Estate














I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Mark B. Rush
Professor of Economics


This dissertation was submitted to the Graduate Faculty of the
Department of Finance, Insurance, and Real Estate in the College of Business
Administration and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy


August 1994 Dean. Graduate School






















1i)











UNIVERSITY OF FLORIDA
Il ll2lll 2 08553 5705llll 111111ll
3 1262 08553 5705